Determination
OF
SulFur Dioxide
Improved Monier-Williams Method JOHN B. THOMPSON AND ELIZABETH TOY Q.M.C. Subsistence Research and Development Laboratory, Chicago Quartermaster Depot, Chicago, 111. A modification of the official A.O.A.C. Monier-Williams method is described which possesses a marked increase in sensitivity over the volumetric application of the official method. Comparisons are made with the official method and a modified Bennett-Donovan method using dehydrated vegetable products.
DURING
recent studies of the effect of time and temperature on sulfur dioxide levels of sulfited cabbage and carrots conducted in the Subsistence Research and Development Laboratory, it w&snecessary to utilize a highly sensitive and reproducible method for determining small sulfur dioxide losses. Although the Monier-Williams method (1) has been adopted as official by the A.O.A.C., it was not found sufficiently sensitive for the authors’ purpose. Methods involving distillation into iodine ( 4 ) were not considered satisfactory, onTing to the volatile sulfur content of cabbage, as they gave abnormally high values on some unsulfited controls. The modification of the Bennett and Donovan procedure as described by Prater, Johnson, Pool, and hlackinney (6) gave rise to considerable difficulty. It was thought that this could undoubtedly be attributed, not to the specificity of the method, but to the conditions under which it was operated in this laboratory. The samples were analyzed every 2 to 4 weeks, and owing to limited personnel, it was not possible to have the same analyst make the determinations each time. As a result, there wra usually some uncertainty of end points, particularly the flash end point. A more serious difficulty was the darkening of the vegetables which occurred under the higher storage temperatures and markedly intepfered with the observance of the end point. When the Bennett and Donovan method was applied to samples withdrawn periodically from high-temperature storage there were often noted increases and losses in sulfur dioxide levels which could not be accounted for by normal variation within a sample, and over-all trends were difficult t o obtain. It is indicated that the authors did not consider the applicability of the method to samples damaged by heat during processing or samples that had been held under storage a t elevated temperatures, In order to satisfy requirements, it became necessary to modify the Monier-Williams method in such manner that its sensitivity would be increased to the same order aa that of the modified Bennett-Donovan method when the latter was operated under ideal conditions with normal samples. Nissen and Petersen (6) have simplified the official Monier-Williams method in two ways: (1) by neutralizing the 3% hydrogen peroxide t o pH 4.0 with sodium hydroxide and (2) by titrating the absorbed sulfur dioxide to pH 4.0 utilizing a glass electrode. Essentially, this speeds up the preparation of the hydrogen peroxide and eliminates the uncertainty of the bromophenol blue end point. Recently Taylor ( 7 ) reported the method had little advantage over the official method, particularly if gravimetric checks were to be obtained. Obviously, gravimetric checks are too time-consming for rapid analyses. There is probably only one fallacy in the volumetric application of the official method and the modified method of R’issen and Petersen, and that lies in the buffering capacity of the absorbed carbon dioxide in the final solution to be titrated. The carbon dioxide serves as a sweeping agent to pass the liberated sulfur dioxide through the condenser into the hydrogen peroxide. An atmosphere of an inert gas also prevents oxidation of the
sulfur dioxide. By the simple expedient of substituting nitrogen for carbon dioxide the above conditions can be obtained, but as shown in Figure 1 the buffering action of carbon dioxide is eliminated and the sensitivity of the titration is markedly increased by approximately 18 times. The end point becomes 5.8 to 6.0 instead of 4.0. Although the glass electrode hss been used entirely in this laboratory, there is no reason why an indicat.or titration could not be substituted.
612
Figure 1.
Titration Curves
Carbon dloxlda
0..
nlbomn
The same effect can be obtained by carefully boiling away the carbon dioxide prior to the titration. This, however, adds one more step and creates an added danger in that loss from spattering may occur in routine analyses. In addition, extra time ic. required, because the solution to be titrated must be broughq to the working temperature of the electrodes. M E T H O D APPLICATION
APPARATUS. The distillation apparatus, diagrammed in Figure 2, consists of a, a 500-ml. wash bottle for pyrogallol reagent: b, a 500-ml. wash bottle for water; c, a 2-liter, 3-necked d i s tilling flask; d, a reflux condenser. e, a dropping funnel; andf, a receiver consisting of a 125-ml. krlenmeyer flask and a Peligot tube, g. This apparatus is preferred by the authors, but is not mandatory. The one described by the A.O.A.C.( I ) may be used, but is not considered as convenient. Titration assembly utilizing a glass and calomel electrode (a Beckman Model G pH meter employing the long electrodes wae used by the authors). REAGENTS.Hydrogen peroxide, 3%,.prepared by diluting a 30% stock solution (Superoxal) and adjusting to pH 4.0 witb 0.1 N sodium hydroxide. It is not necessary to adjust the strength of this reagent to exactly 3% ( 5 ) . Pyrogallol solution (8). Three hundred grams of pyrogallic acid in 1 liter of water; add 2.5 volumes of 50% sodium hydroxide. Nitrogen, water-pumped, 99.7’3, pure, commercial cylinder
ANALYTICAL EDITION
October, 1945
Hydrochloric acid, concentrated. Sodium hydroxide, standard 0.010, 0.025, 0.050, and 0.10 N solutions. PROCEDURE. Connect the distilling apparatus as shown in Figure 2, using rubber stoppers throughout. Add to. wash bottles a and b about 400 ml. of pyrogallol reagent and distilled water, respectively. The bottles require only infrequent recharmng, as the amount of oxygen present in the nitrogen is very low. Add 15 ml. of the 3% hydrogen peroxide reagent to receiving flask, f, and 5 ml. to the Peligot tube, 8. Connect the receiving flask to the upper end of the condenser, making certain that the delivery tube extends below the surface of the hydrogen peroxide in t,he receiving flask. Sweep out the assembled apparatus for approximately 5 minutes with the oxygen-free nitrogen, adjusting the gas so that there is a steady flow through the Peligot tube. Remove the stopper containing the dropping funnel, add the sample, usually 25 grams, and immediately restopper. Avoid adding in such a manner that some of the sample may remain in the neck of the flask. Add 300 ml. of recently boiled and still hot distilled water, containing 10 ml. of hydrochloric acid, through the dropping funoel. The addition must be made steadily and a small portion should be allowed to remain in the dropping funnel to avoid loss of Liberated sulfur dioxide through back pressure. Reflux for 1 hour (in the case of dehydrated vegetables) in a current of nitrogene Shut Off the water in the condenser and continue the distillation until the delivery tube from the condenver becomes just warm (this step may be omitted in the case of dehydrated vegetables). Transfer the contents of the receiving flask and Peligot tube to a 250-ml. beaker with adequate washing. Dilute to aPProximatelY 100 ml., and titrate with standard alkali t o PH 6.0. . Choose the normality on the basis of the suspected sulfur dioxide content or the initial pH, Table I gives the approximate initial pH values cciresponding to the normalities of the sodium hydroxide to be used. Correct for a reagent blank run in identical manner and calculate as p.p.m. of sulfur dioxide.
Table 1. Normality of Sodium Hydroxide to Be Used (On ha& of initial pH of solution to be titrated) Initial Normality Mg.of €303 Equivalent t o PH of NaOH 1 MI. of Standard NaOH 2.6 0.010 0.32
Table
613
II. Recovery of Sulfur Dioxide A d d e d to Blank Determinations (Modified Monier-Williams method)
so:,
SO¶
Present
Reoovered
Recovery
MQ.
MQ.
%
91.12 91.30 92.26 90.10 88.54 88.54 20.86 21.09 8.32 8.22 3.97 3.96
90.21 90.80 92.35 88.49 88.16 88.19 21.35 21.10 8.30 8.25 3.95 3.90
99.o 99.5 100.1 96.1 99.6 99.6 102.4 100.1 99.8 100.4 99.5 98.5 b
DISCUSSION
The recovery of added sulfur dioxide t o aqueous solutions or blank determinations is shown in Table 11. Sodium metabisulfite solutions stabi,bilized with sodium pyrophosphate decahyRecovery Of drate were wed' The recovery was added sulfur dioxide t o cabbage and carrots is shown in Table 111. The recovery data are comparable with those reported by Prater and co-workers for the modified Bennett-Donovan method in the addition Of (6). It is thought that the extra step the sulfur dioxide solution, following the addition of the sample, may partially account for the lower results obtained when compared to those previously shown for the aqueous solutions. Although it has been reported (3) that sulfhydryls do not interfere in the Monier-Williams method, it was thought advisable to check it a,~ a possible, but not probable, source of error. Determinations run with added hydrogen sulfide gave titration curves which coincided with those without added hydrogen sulfide. A blank consisting of water saturated with hydrogen sulfide gave a zero titration with 0.01 N sodium hydroxide. In the latter case sulfur appeared t o be precipitated to some extent in the hydrogen peroxide, but this did not interfere. To study the effect of varying sample size, a series of 5-, IO-, 15-, 25-, and 50-gram samples were prepared. The samples were ground in the Wiley Mill with a 1.0-mm. screen, utilizing small charges to avoid loss of sulfur dioxide due t o heat, until practically all had passed through the screen. Care was taken to blend the charges well. Prepared samples were stored under refrigeration. These precautions were taken in order t o avoid the errors due to the possibility of uneven sulfiting and nonuniform particle size, and to ensure reproducibility of results. The results shown in Table I V indicate that on the basis of the calculated coefficients of variation (percentage of relative variation of the distribution about the mean) utilization of 25- to 6G
Table 111.
Recovery of Sulfur Dioxide A d d e d to Dehydrated Cabbage and Carrots (Modified Monier-Williams method)
SO¶
Originally Preeent Ma.
18.96 18.96 18.96 18.96 18.96 22.22 22.80 22.80 22.77
Figure 2.
Distillation Apparatus
so:,
Added
MQ.
Son
Found Ma.
so1
Recovered Ma.
14.82 19.13 19.90 19.65 9.44 20.22
Cabbage 33.70 37.46 37.81 37.74 28.46 41.79
14.74 18.50 18.85 18.78 9.50 19.57
19.49 9.82 45.70
Carrots 42.05 32.26 66.84
19.25 9.46 44.07
Recovery
% 99.5 96.7 94.7 95.6 100.6
96.8 98.8 96.3 96.4
614
INDUSTRIAL AND ENGINEERING CHEMISTRY
Table
IV. Effect of Vsr in Sample Sizes in Determination of 4ul!ur Detn. No.
1 2 3 4 6 6
... ... ... ... .... ..
Cabbage 752 727 691 783 723 751 750 749 760 773 155 767 1.7 2.8
1 2
924 924 929 934 915 925 0.6
Carrots 909 905 916 90 1 925 903 0.9
a
C.Y., % b
a 4 6 6
0
Dioxide
(Modified Monier-Williams method) Sample Weight 50g. 25 g. 15g. log. P a m . of Sulfur Dwzids
... ... ... ... ... ... ...
5 g.
770 789 665 803 672 707 7.6
776 83 1 727 783 816 716 5.3
... ... ...
... ... ... ... ... ... ...
... ... ...
C.V., % ... In the case. of dehydrated cabbage di5culty waa encountered when
rample size waa increaaed much in exceaa of 25 grama owing to charring in Baak during dintillation.
where
C.V.
-
coe5cient of variation in er cent d = deviation of individual Jeberminatiom from the mean, r e gardless of sign n number of determinations M = mean
gram samples offers the best opportunity of reproducing results. If a large number of samples were run at each level, it is highly probable that the average values would be about the same. The modified method was compared with both the official and the modified Bennett-Donovan method when used for cabbage and carrots. Table V summarizes comparative data obtained from 37 determinations on a single sample of cabbage and 27 determinations on a single sample of carrots by the three methods, Of the three methods the official A.O.A.C. gave the poorest replication, but consistency of replication was very good for either the modified Monier-Williams or the modified Bennett-Donovan method. The mean value for cabbage obtained by the former is only slightly lower than that found for the latter, while the mean value found for carrots was substantially higher. It was thought that possibly the pigmentation of carrots caused an overtitration of the so-called “flash” end point which would tend t o yield lower results by the Bennett-Donovan procedure. The difference, however, was too great to make this assumption entirely logical. I n the absence of unsulfited controls a further check was made on fnxhly ground raw carrots. Samples were taken of such size to be comparable to 25 grams, in the case of the Monier-Williams methods, and 8 grams, in the case of the Bennett-Donovan method, of 10% moisture producte. No attempt was made to remove any of the volatile materials which would be liberated during dehydration. No sulfur dioxide was found by either of the Monier-Williams methods and *30 p.p.m. were found by the Bennett-Donovan method. Nevertheless, because of the incomplete binding of the acetone and eulfur dioxide in the Bennetb Donovan method as the titration approaches completion, it was suggested (8) that glyoxal should be used in place of the acetone. Four milliliters of a 30% solution were found sufficient to bind the sulfur dioxide, and the end point was more persistent and much easier to detect. However, the same sample of carrots yielded similar low results comparable to those obtained when acetone was used, and it was postulated that this particular sample had been heat-damaged in processing.
checked by the modified Bennett-Donovan method. Generally good agreement was obtained in all three samples (Table VI). This further emphasizes that the Bennett-Donovan method may not be satisfactory when applied to heat-damaged samples. The data presented in Table VI may be utilized in explaining the data in Table V for carrots. It has been shown (Table VI) that good agreement is obtained between the modified MonierWilliams and the Bennett-Donovan procedures when normal sulfited carrots are used. It can be anticipated that all three methods would agree on normal samples. The amount of sulfur dioxide absorbed by hydrogen peroxide would not vary on merely changing the carrier gas, as this gas does not enter into the chemical reaction. Although it would appear from Table V that the mean values for carrots when obtained by the Bennett-Donovan and the official Monier-Williams methods are in agreement, it is suggwted that this is not the case and that if a large number of determinations had been made by the official method the mean value would have been in agreement with that obtained by the modified Monier-Williams procedure. Potentiometric curves have been plotted for a series of routine samples, including both those which were normal and obviously scorched or heat-damaged. In no case has there been noted any deviation from the typical curve shown in Figure 1. Any change in the shape of the curve would indicate the presence of substances other than sulfur dioxide in the hydrogen peroxide. This is another indication of the specificity of the method. SUMMARY
The official A.O.A.C. Monier-Williams method for the determination of sulfur dioxide, though basically correct, is insufficiently sensitive. Its sensitivity may be increased approximately 18 times by eliminating the buffering action of the carbon dioxide in the solution to be titrated. This may be achieved by either employing oxygen-free nitrogen as the carrier gas or eliminating the dissolved carbon dioxide by careful boiling. A modification of the official Monier-Williams method utilizing oxygenfree nitrogen as a carrier gas and potentiometric titration may be applied equally as well to samples of low or high sulfur dioxide
Table V.
were analyzed by the modified Monier-Williams procedure and
Comparison of Three Methods for Determination of Sulfur Dioxide Monier-Williams Modified
Monier-Williams A.O.A.C.
Bennett-Donovan
Cabbage
Range, p.p.m. C.V.. %
M
2.3
9.9
5.8
CarrotI
No. of detns. Max., p.p.m. Min., p.p.m. Mean, p.p.m. Range C.V.,3 . P . m ‘
10 926 901 911 M * 13 0.9
Table VI. Sample No. 1 2 3 0
7 891 545 690 M * 173 22.5
10 730 650 691 M * 40 4.3
Data on Normal Sulfited Carrots
SOr by Modified
In support of this assumption three samples of dehydrated carrots, in which heat damage or scorch could not be detected,
Vol. 17, No. 10
Monier-Williama Method P.p.m. 664 653 1469 1471 197 196
SO: by Modified BennettDonovan Method Ueing Uaing acetone nlvoxal -P.p.m. P.p.m. 600 610
Titration of lens than 1 ml. of 0.05 N iodine.
1406 1414 1220 96
1422 1414 102 130
ANALYTICAL EDITION
October, 1945
61s
LITERATURE CITED
Mackinney, Gordon, University of California, Division of Food Technology, Berkeley, Calif., private communication. Monier-Williams, G. W., Reports on Public Health and Medical Subjects, No. 43. London Ministry of Health, 1927. Nichols, P. F., and Reed, H. M., IND.ENO.CHEM., ANAL.ED.,4, 79 (1932). Nissen, B. H., and Petersen, R. B., Ibid., 15,129 (1943). Prater, A. N., Johnson, C. M., Pool, M. F., and Mackinney, G., Ibid., 16,153 (1944). (7) Taylor, L. V., J.‘A.uoC. Oficial Agr. Chem., 27,386 (1944). (8) Townley, R.C., and Gould, I. A., J. Dairy Sci., 26,689 (1943).
(1) AsSoc. official Agr. Chem., OfLicial and Tentative Methods of Analysis, 5th ed., pp. 463-4 (1940).
PREEENTED at the Conference on Recent Technical Advances in Army Dehydrated Foods, Chicago, Feb. l, 1946.
content. It utilizes a fairly large sample size, appears to be specific, and is unaffected by pigments, pigment changes, or sulfhydryls. ACKNOWLEDGMENT
Grateful acknowledgment is made to Gordon Mackinney, University of California, Division of Food Technology, for submission of the three samples on which the data in Table VI are obtained
.
Polarographic Analysis of Aluminum Alloys 1. M. KOLTHOFF AND GEORGE MATSUYAMA School of Chemistry, University of Minnesota, Minneapolis, Minn.
Procedures have been developed for the polarographic determination of iron, copper, lead, nickel, and zinc in aluminum alloys. The alloy i s heated with sodium hydroxide and the solution completed in nitric acid. In the absence of chloride, ferric iron and copper give well separated waves. If the ratio of iron to copper i s large, the ferric iron i s reduced with hydroxylamine hydrochloride. The lead wave i s determined after reduction of the ferric iron and precipitation of copper as cuprous thiocyanate and adjustment of the p H . The nickel and zinc waves are determined after adjustment of the p H of the solution of the alloy, and addition of hydroxylamine hydrochloride, thiocyanate, sodium citrate, and pyridine. Alternate procedures are discussed. The proposed procedures give satisfactory results and are especially recommended in routine analysis. The actual time spent in the total analysis for the five elements will be less than 45 minutes in routine work.
IN
THIS paper are discussed rapid, reliable methods for the polarographic analysis of iron, copper, lead, nickel, and zinc in aluminum alloys. The methods developed for iron, copper, and lead are both rapid and accurate. These metals can be determined with ordinary polarographic accuracy in aluminum alloys by the procedure recommended. The method developed for nickel can be made more accurate by more or less elaborate separations from aluminum and other metals, but the recommended method is sufficiently accurate for routine analyses. Two methods are described for zinc; one gives rapid results but is not so accurate as the other, which is more time-consuming. Since most of the analyses are made in the presence of aluminum, the ‘‘supporting electrolyte” is usually high in aluminum content. The diffusion current constant of the various metals depends on the concentration of aluminum and must be determined in a supporting electrolyte containing approximately the same amount of aluminum as the unknown. The aluminum content can be varied a t least 5% without affecting the diffusion current constant. If, in the final analysis, the aluminum content varies greatly from the usual concentration, it is necessary to adjust the aluminum content of the unknown solution or to determine the diffusion current constant in this solution by adding a known amount of the element and again determining the diffusion current. I n the present work, a uniform method for the dissolution of the alloy is used. Stated briefly, this method involves treatment with 20% sodium hydroxide solution to dissolve the aluminum, followed by treatment with 1 to 1 nitric acid (specific gravity 1.22). Hydrochloric acid is avoided, since it interferes in the determination of iron and copper. Chloride ion gives an anodic
wave beginning a t around 0 volt (us. S.C.E.), so that the iron wave, shifted to this potential, coincides with the copper wave in solutions containing chloride. Synthetic aluminum alloy solutions were prepared from solutions of the following: 99.9% pure aluminum metal obtained from the Aluminum Company of America, Baker and Adamson nickel shot, the solution of which was analyzed for nickel by precipitation with dimethylglyoxime, Baker standardizing zinc foil Baker standardizing iron wire, Baker electrolytic copper foil, and Merck reagent lead nitrate. The aluminum solution was prepared by the same procedure as employed with alloys (Section 2). The other metals were dissolved in a small volume of 1 to 1 nitric acid and diluted with water t o make solutions containing 1.00 mg. per ml. or other appropriate amounts. Polarographic measurements were made with both a SargentHeyrovskS. Model XI polarograph and a manual apparatus. Oxygen was removed from the polarographic solutions by bubbling with nitrogen gas which was purified by passing through solutions of chromous chloride, sodium hydroxide, and water. The temperature used in all this work was 25.0’ C. The chemicals used were of reagent grade or better. The polarographic cells were of the type described by Hume and Harris (1). Tube D has a 1-cm. fine sintered-glass disk a t the end. The agar bridge, C , should be inserted into tube D just before the polarogram is run to prevent the diffusion of potassium chloride from the agar bridge into the polarographic solution. This precaution is important in the determination of iron, since chloride interferes in the separation of the polarographic waves of iron and copper. All potentials in this paper refer to the saturated calomel electrode. Polarographic Determination of Iron, Copper, Lead, Nickel, and Zinc in Aluminum Alloys 1. SOLUTIONSREQUIRED.Bromocresol Green (0.1%). To 100 mg. of bromocresol green (tetrabromo-m-cresokiulfonphthalein) in an agate mortar add 2.9 ml. of 0.05 A4 sodium hydroxide. Rub the solid until dissolved and dilute with water to 100 ml. Dithizone (0.05% solution in carbon tetrachloride). Diawlve 500 mg. of dithizone (diphenylthiocarbazone) in 1000 ml. of carbon tetrachloride (U.S.P.). Gelatin (0.5%). Soak 0.5 gram of gelatin in 100 grams of distilled water. Heat to the boiling point, cool to room temperature, and add a few drops of toluene as a presemative. Hydrochloric Acid (0.1 M ) . Dilute 8.5 ml. of constant-boiling acid to 500 ml. Hydroxylamine Hydrochloride (2 M ) . To 13.9 grams of hydroxylamine hydrochloride (recrystallized from water) add distilled water to make 100 ml. Kitric Acid (1 to 1 or sp. gr. 1.22). Mix 500 ml. of concentrated nitric acid (sp. gr. 1.42) with 500 ml. of distilled water. Potassium Thiocyanate (2 M ) . To 19.4 grams of reagent potassium thiocyanate, add distilled water to make 100 ml. of solution. Sodium Citrate (1.25 M ) . To 184 grams of reagent sodium citrate dihydrate add distilled water to make 500 ml. of solution.
